Analysis Relative Ratings: An exercise of comparing stats in context

[Engimal v3]

The Premise
You’re almost certain to have observed a post on BigFooty along these lines:

“Ruckman X got 3 more hitouts than Ruckman Y this round, in two completely separate games, so clearly they are better!”

It’s never sat right with me that people do this, without considering that (for example) a game might be higher scoring, meaning centre stoppages are more common resulting in more opportunities for hitouts.

If we think about goals, are two players kicking 2 goals really the same if one is in a high-paced, high-scoring game, and the other is coached by Ross Lyon?

Of course not, right? So, it got me thinking: What if I were able to obtain second-by-second stats from a game and then be able to rank those stats in relation to what other players achieved in that quarter?

I mean 10 disposals in a quarter is impressive, but what if there’s other players with 15 or 20? It’s less impressive relative to the flow of that particular game. Similarly, kicking 1 goal in a wet game is impressive, but not that impressive in a shootout.



The Execution
Through the power of prayer and positive thinking, I obtained this data and built a system:

• For each quarter of a game, each players’ stats will be recorded
• Each stat (e.g. goals, hitouts to advantage, contested posessions, score launches, etc) will be normalised by assessing the minimum/maximum of that statistic in the quarter
• All stats counts for a quarter will therefore be on a common scale between 0 and 1
• For each quarter, the player gets ranked for that stat on the following scale:

[TD valign="bottom"] CATEGORY [/TD][TD valign="bottom"] MINIMUM SCORE [/TD][TD valign="bottom"] MAXIMUM SCORE [/TD]

[TD valign="bottom"]EXCELLENT[/TD]
[TD valign="bottom"]

0.8​

[/TD]
[TD valign="bottom"]

1.0​

[/TD]

[TD valign="bottom"]GOOD[/TD]
[TD valign="bottom"]

0.6​

[/TD]
[TD valign="bottom"]

0.8​

[/TD]

[TD valign="bottom"]FAIR[/TD]
[TD valign="bottom"]

0.4​

[/TD]
[TD valign="bottom"]

0.6​

[/TD]

[TD valign="bottom"]POOR[/TD]
[TD valign="bottom"]

0.2​

[/TD]
[TD valign="bottom"]

0.5​

[/TD]

[TD valign="bottom"]BAD[/TD]
[TD valign="bottom"]

0.0​

[/TD]
[TD valign="bottom"]

0.2​

[/TD]

From there, I then took each player and mapped out the % of quarters in EXCELLENT/GOOD rating


The Insights

Below are some charts for some selected stats from 2021, to illustrate the results. For all the below insights, a limit was set that at least 53 quarters had to be played throughout 2021, which was the median across players according to the data I had. The full set of data is much larger, but a lot of it is not particularly interesting.

Hitouts to Advantage

• Hitout machine Nic Naitanui had 82.14% of his 2021 Quarters rated as Excellent/Good
• This chart highlights the elite ruckmen of the competion, with the first 9 ruckmen rated as Excellent/Good in at least 50% of quarters, with a sharp decline with 10th placed Rhys Stanley
  1

  Rhys Stanley

  Age

  33

  Ht

  200cm

  Wt

  102kg

  Pos.

  Ruck

  Career

  Season

  Last 5

  D

  13.6

  3star

  K

  7.2

  3star

  HB

  6.4

  4star

  CL

  3.4

  5star

  HO

  27.6

  5star

  D

  12.0

  3star

  K

  7.5

  3star

  HB

  4.5

  3star

  CL

  5.0

  5star

  HO

  21.5

  5star

  D

  11.8

  3star

  K

  5.4

  2star

  HB

  6.4

  4star

  CL

  3.0

  5star

  HO

  33.0

  5star

Contested Intercept Marks

• Despite averaging an intercept mark less than intercept average leader Jake Lever
  8

  Jake Lever

  Age

  28

  Ht

  194cm

  Wt

  94kg

  Pos.

  Def

  Career

  Season

  Last 5

  D

  14.6

  4star

  K

  9.7

  4star

  HB

  4.9

  4star

  M

  5.4

  5star

  T

  1.3

  3star

  MG

  242.0

  4star

  D

  9.6

  2star

  K

  6.5

  3star

  HB

  3.1

  2star

  M

  3.9

  4star

  T

  1.3

  2star

  MG

  145.4

  3star

  D

  16.0

  4star

  K

  11.0

  4star

  HB

  5.0

  4star

  M

  6.6

  5star

  T

  1.6

  4star

  MG

  189.8

  4star

  , Jeremy McGovern
  20

  Jeremy Mcgovern

  Age

  32

  Ht

  197cm

  Wt

  97kg

  Pos.

  Def

  Career

  Season

  Last 5

  D

  14.7

  4star

  K

  10.1

  4star

  HB

  4.6

  3star

  M

  6.8

  5star

  T

  1.5

  3star

  MG

  238.7

  4star

  D

  12.4

  3star

  K

  8.6

  3star

  HB

  3.8

  3star

  M

  6.4

  5star

  T

  2.0

  3star

  MG

  183.4

  3star

  D

  11.0

  3star

  K

  8.8

  4star

  HB

  2.2

  2star

  M

  7.2

  5star

  T

  1.4

  4star

  ranks #1 in contested intercept marks
  • He ranked EXCELLENT/GOOD in over a third of his 2021 quarters

Frees For

• Luke Parker has the 7th most toal frees in 2021, yet he ranks #53 (not pictured) in the Relative Ratings
  • In only 11 quarters of Parker's season was he rated as a Excellent/Good free-receiver in the quarter.
    • Marcus Bontempelli (received 2 more frees in 2021), was rated Excellent/Good in 19 quarters, for comparison.
• In the other direction, Brodie Grundy
  4

  Brodie Grundy

  Age

  30

  Ht

  202cm

  Wt

  105kg

  Pos.

  Ruck

  Career

  Season

  Last 5

  D

  17.5

  4star

  K

  8.0

  3star

  HB

  9.6

  5star

  CL

  4.1

  5star

  HO

  32.0

  5star

  D

  16.4

  4star

  K

  6.8

  3star

  HB

  9.6

  5star

  CL

  4.0

  5star

  HO

  34.6

  5star

  D

  13.8

  4star

  K

  5.2

  2star

  HB

  8.6

  5star

  CL

  2.6

  5star

  HO

  20.6

  5star

  barely sits inside the top 25 players for total Frees For, and yet he places 8th in the Relative Ratings

Goals

• Matt Taberner, despite coming 20th in the Coleman, jumps to 4th place in the Relative Ratings
  • 35% of the Fremantle forward's 2021 quarters were rated as Excellent/Good
• Tom Hawkins kicked more 2021 goals than Franklin, Greene, Tabener, and King, yet places behind all in the Relative Ratings
  • ~5% fewer of Hawkins' quarters were viewed as Excellent/Good than Franklin's

The… other part where I acknowledge shortcomings of the system and end the post

At the end of the day, I'm a footy enthusiate who works in software dev, but I'm not a statician or a trained data scientist. Take any of the "insights" above with a grain of salt, and at most something to think about when shitfighting in All-Australian threads. These games don't exist in a controlled environment. Australian Rules Football by its design is a 360 degree field of chaos. It has an incredible amount of variables that can influence the pace of football played (e.g. weather), the way certain stats are gained (e.g. high-stoppage vs low-stoppage games), and even the way a game is officiated. The difference between two players in a game or a season can not be determined by a Footywire player comparison link.

I did this mostly for me because I've gone a bit stir-crazy in lockdown, but maybe others will find something interest in it.

 

[Ambrosia]

I’m not sure I like this method for goals not that I have any intelligent justification off the top of my head.

The ruck example reminds me of Will Minson

Will Minson

Age

39

Ht

199cm

Wt

106kg

Pos.

Ruck

Career

Season

Last 5

D

11.2

3star

K

6.0

2star

HB

5.2

4star

CL

3.5

5star

HO

21.3

5star

No current season stats available

D

3.4

1star

K

2.0

1star

HB

1.4

2star

HO

7.4

5star

’s dud AA selection. He had good totals because the Dogs played a really high stoppage game plan. There’s no way he should have gotten AA ahead of I think Goldstein that year.

 

[Engimal v3]

I’m not sure I like this method for goals not that I have any intelligent justification off the top of my head.

The ruck example reminds me of Will Minson’s dud AA selection. He had good totals because the Dogs played a really high stoppage game plan. There’s no way he should have gotten AA ahead of I think Goldstein that year.

It's definitely more useful for some stats over others. I still think it works with goals, overall. Tim Taranto kicking a goal in a 12 goal quarter in Round 14 is rated as Fair where Peter Wright's goal in a 2 goal quarter in Round 17 is Excellent. And fair enough, Wright kicked half the goals for the quarter, where Taranto snuck a goal in during a goal bonanza.

 

[Some Idiot]

I’m not sure I like this method for goals not that I have any intelligent justification off the top of my head.

The ruck example reminds me of Will Minson’s dud AA selection. He had good totals because the Dogs played a really high stoppage game plan. There’s no way he should have gotten AA ahead of I think Goldstein that year.

Bulldogs vs kangaroos Minson had more clearances and more hitouts than Goldstein. No idea what the hitouts to advantage was like. Acknowledge that Goldstein was a bit better around the ground but Minson had a fantastic season as a tap ruckman.

 

[Kwality]

Not intended as a criticism, more a clarification of what is an intercept mark as distinct from a mark in the back half. My understanding originally its where a backman left his man & backed himself to intercept an attack. In the last half of 2021 it seemed to be a backman taking any mark.
What is it here?

& good work

 

[Engimal v3]

Not intended as a criticism, more a clarification of what is an intercept mark as distinct from a mark in the back half. My understanding originally its where a backman left his man & backed himself to intercept an attack. In the last half of 2021 it seemed to be a backman taking any mark.
What is it here?

& good work

Intercept mark can be taken anywhere, just has to be from an opposition kick

 

[00Stinger]

...

Sent from my SM-G981B using Tapatalk

 

[Ambrosia]

Bulldogs vs kangaroos Minson had more clearances and more hitouts than Goldstein. No idea what the hitouts to advantage was like. Acknowledge that Goldstein was a bit better around the ground but Minson had a fantastic season as a tap ruckman.

Yes the bulldogs were a really high stoppage team which inflated Minson’s stats

 

[KissKiss]

It's definitely more useful for some stats over others. I still think it works with goals, overall. Tim Taranto kicking a goal in a 12 goal quarter in Round 14 is rated as Fair where Peter Wright's goal in a 2 goal quarter in Round 17 is Excellent. And fair enough, Wright kicked half the goals for the quarter, where Taranto snuck a goal in during a goal bonanza.

It really is an interesting thing to consider with different stats and different game styles. Ie the Lions are a low possession-high kick team (kick it forward as quickly as possible is mostly the plan) versus the dogs who are high possession-high handballs (sharing the ball around more before moving it forward). So a guy like Jarrod Lyons (average 28 touches) is often rated a bit below other inside mids (ie Mitchell 34, Macrae 34, Laird 32, Parish 30). Is he a bit below them or are we not looking at things close enough to understand them